Translate: Don't filter three-letters language code

1) Create IsValidLanguageCode instead of ResetLanguageCode because I wanted to make the logic clearer.
2) Implement IsValidLanguageCode not to filter three-letters language code like 'ceb'.

BUG=256536
TEST=unit_tests --gtest_filter=TranslateHelperTest.*

Review URL: https://chromiumcodereview.appspot.com/18135003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@209923 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 59 insertions, 28 deletions

diff --git a/chrome/renderer/translate/translate_helper.cc b/chrome/renderer/translate/translate_helper.cc
index 34b4afe..fbc4d06 100644
--- a/chrome/renderer/translate/translate_helper.cc
+++ b/chrome/renderer/translate/translate_helper.cc
@@ -9,6 +9,7 @@
 #include "base/logging.h"
 #include "base/message_loop.h"
 #include "base/strings/string16.h"
+#include "base/strings/string_split.h"
 #include "base/strings/string_util.h"
 #include "base/strings/utf_string_conversions.h"
 #include "chrome/common/chrome_constants.h"
@@ -327,14 +328,41 @@ void TranslateHelper::CorrectLanguageCodeTypo(std::string* code) {
 }
 
 // static
-void TranslateHelper::ResetInvalidLanguageCode(std::string* code) {
-  // Roughly check if the language code follows [a-z][a-z](-[A-Z][A-Z]).
-  size_t dash_index = code->find('-');
-  if (!(dash_index == 2 && code->size() == 5) &&
-      !(dash_index == std::string::npos && code->size() == 2)) {
-    // Reset |language| to ignore the invalid code.
-    *code = std::string();
+bool TranslateHelper::IsValidLanguageCode(const std::string& code) {
+  // Roughly check if the language code follows /[a-zA-Z]{2,3}(-[a-zA-Z]{2})?/.
+  // TODO(hajimehoshi): How about es-419, which is used as an Accept language?
+  std::vector<std::string> chunks;
+  base::SplitString(code, '-', &chunks);
+
+  if (chunks.size() < 1 || 2 < chunks.size())
+    return false;
+
+  const std::string& main_code = chunks[0];
+
+  if (main_code.size() < 1 || 3 < main_code.size())
+    return false;
+
+  for (std::string::const_iterator it = main_code.begin();
+       it != main_code.end(); ++it) {
+    if (!IsAsciiAlpha(*it))
+      return false;
   }
+
+  if (chunks.size() == 1)
+    return true;
+
+  const std::string& sub_code = chunks[1];
+
+  if (sub_code.size() != 2)
+    return false;
+
+  for (std::string::const_iterator it = sub_code.begin();
+       it != sub_code.end(); ++it) {
+    if (!IsAsciiAlpha(*it))
+      return false;
+  }
+
+  return true;
 }
 
 // static
@@ -342,11 +370,12 @@ void TranslateHelper::ApplyLanguageCodeCorrection(std::string* code) {
   // Correct well-known format errors.
   CorrectLanguageCodeTypo(code);
 
-  // Convert language code synonym firstly because sometime synonym code is in
-  // invalid format, e.g. 'fil'. After validation, such a 3 characters language
-  // gets converted to an empty string.
+  if (!IsValidLanguageCode(*code)) {
+    *code = std::string();
+    return;
+  }
+
   TranslateUtil::ToTranslateLanguageSynonym(code);
-  ResetInvalidLanguageCode(code);
 }
 
 // static
diff --git a/chrome/renderer/translate/translate_helper.h b/chrome/renderer/translate/translate_helper.h
index abef6ca..21fbcaa 100644
--- a/chrome/renderer/translate/translate_helper.h
+++ b/chrome/renderer/translate/translate_helper.h
@@ -87,6 +87,7 @@ class TranslateHelper : public content::RenderViewObserver {
   virtual double ExecuteScriptAndGetDoubleResult(const std::string& script);
 
  private:
+  FRIEND_TEST_ALL_PREFIXES(TranslateHelperTest, IsValidLanguageCode);
   FRIEND_TEST_ALL_PREFIXES(TranslateHelperTest, AdoptHtmlLang);
   FRIEND_TEST_ALL_PREFIXES(TranslateHelperTest,
                            CLDAgreeWithLanguageCodeHavingCountryCode);
@@ -106,8 +107,8 @@ class TranslateHelper : public content::RenderViewObserver {
   // Converts language code to the one used in server supporting list.
   static void ConvertLanguageCodeSynonym(std::string* code);
 
-  // Resets language code if the specified string is apparently invalid.
-  static void ResetInvalidLanguageCode(std::string* code);
+  // Checks if the language code's format is valid.
+  static bool IsValidLanguageCode(const std::string& code);
 
   // Applies a series of language code modification in proper order.
   static void ApplyLanguageCodeCorrection(std::string* code);
diff --git a/chrome/renderer/translate/translate_helper_unittest.cc b/chrome/renderer/translate/translate_helper_unittest.cc
index e1024ee..88f3b1f 100644
--- a/chrome/renderer/translate/translate_helper_unittest.cc
+++ b/chrome/renderer/translate/translate_helper_unittest.cc
@@ -30,32 +30,33 @@ TEST_F(TranslateHelperTest, LanguageCodeTypoCorrection) {
   EXPECT_EQ("ja-JP", language);
 }
 
-// Tests that invalid language code is reset to empty string.
-TEST_F(TranslateHelperTest, ResetInvalidLanguageCode) {
+// Tests if the language codes' format is invalid.
+TEST_F(TranslateHelperTest, IsValidLanguageCode) {
   std::string language;
 
   language = std::string("ja");
-  TranslateHelper::ResetInvalidLanguageCode(&language);
-  EXPECT_EQ("ja", language);
+  EXPECT_TRUE(TranslateHelper::IsValidLanguageCode(language));
 
   language = std::string("ja-JP");
-  TranslateHelper::ResetInvalidLanguageCode(&language);
-  EXPECT_EQ("ja-JP", language);
+  EXPECT_TRUE(TranslateHelper::IsValidLanguageCode(language));
+
+  language = std::string("ceb");
+  EXPECT_TRUE(TranslateHelper::IsValidLanguageCode(language));
+
+  language = std::string("ceb-XX");
+  EXPECT_TRUE(TranslateHelper::IsValidLanguageCode(language));
 
-  // Invalid because of three characters before hyphen.
+  // Invalid because the sub code consists of a number.
   language = std::string("utf-8");
-  TranslateHelper::ResetInvalidLanguageCode(&language);
-  EXPECT_TRUE(language.empty());
+  EXPECT_FALSE(TranslateHelper::IsValidLanguageCode(language));
 
   // Invalid because of six characters after hyphen.
   language = std::string("ja-YUKARI");
-  TranslateHelper::ResetInvalidLanguageCode(&language);
-  EXPECT_TRUE(language.empty());
+  EXPECT_FALSE(TranslateHelper::IsValidLanguageCode(language));
 
-  // Invalid because of three characters.
-  language = std::string("YMO");
-  TranslateHelper::ResetInvalidLanguageCode(&language);
-  EXPECT_TRUE(language.empty());
+  // Invalid because of four characters.
+  language = std::string("DHMO");
+  EXPECT_FALSE(TranslateHelper::IsValidLanguageCode(language));
 }
 
 // Tests that similar language table works.